
This study examines how flag states utilize exemptions and equivalents within the International Maritime Organization’s regulatory system using ten years of cleaned and enriched GISIS notifications. Data preparation used a deterministic, rule-based pipeline that transformed semi-structured text through tokenization, regex patterning, and n-gram co-occurrence analysis. Standardized instrument references and reconstructed analytical fields were linked to ship-type and size metadata. Analyses covered exemption intensity, cross-flag variation, convention portfolios, ship-type associations, temporal dynamics, and exploratory clustering. Results shows that exemptions concentrate in technical regimes led by SOLAS, followed by MARPOL, COLREG, and ICLL; national portfolios are coherent rather than random; and volumes increase over the decade despite exposure adjustment. Notification rates remain highly right-skewed, with a small subset of flags accounting for a disproportionate share of documented exemptions and equivalents. Methodologically, the paper contributes a reproducible, deterministic pipeline that normalizes instrument references, reconstructs exemption fields, and links notifications to vessel attributes, enabling exposure-adjusted cross-flag comparisons and portfolio analysis. The scope is limited to convention-level categories rather than clause- or rule-specific analysis because of fragmented and heterogeneous citation patterns in flag-state reports. Findings have implications for transparency, comparability, and the design of structured reporting in GISIS to support evidence-based oversight.
Keywords: IMO Regulations, Exemptions, rule-based pattern mining